Testing 904a...

[rarecomponent]

At "Step 3 " of the 904a test procedure it says that voltages of R. 35 and R39 should read about 1.0 volts.
Well the reading from my clone 904a at these test points is nothing like. In fact I'm getting minus voltages. If I adjust R40 to full rotation I get a 1 volt at R39 but R35"_is still minus voltage.

Is this test point crucial ? I'm using the actual track PCBs of the original 904 a's.

My concern is the the filter does not self oscillate and seems to lack
Resonance . I have the suggested 1.2 and 2.2 k piggy back at R11.

Apart from this the filter works fine.

Anyone any knowledge of this ?

[analogmonster]

At this point the service manual is not precise enough. There are 3 possibilities of measuring voltages at resistors:

1.) Measuring resistor input vs. ground. In this case these points are the collectors of Q19 and Q20. In my clone I measured a value of 0.6 volts at both resistors.

2.) Measuring resistor output vs. ground. In my clone I had values of -5.4 V at both resistor outputs.

3.) Voltage drop about the resistors. Well, just the difference of both, or to be more precise, the absolute value of that.

So I assumue the first point is meant here. But the main issue here is equality of the voltages anyway, not the absolute value.

My clone works fine in any function with the values measured above, see also http://www.analog-monster.de/mmt904a_en.html

[CZ Rider]

One difference I see in the schematics is the original 7/25/67 schematic (Rev 0) drawing #1149, at the top of the output buffer section, the +12 volts connects to both R44(1) 10 ohm and R42 330 ohm. In the latter revision it is drawn with the +12 volt only connecting to the 10 ohm then connecting to the 330? There are also two CL-1 diodes connected between the -6 volt rail, the 10 ohm resistor, and that same test point at Q19 ? My 904A does not have those two extra CL-1's. There are a few other differences too.
The August 1969 built 904A is a revision "A", and it differs in the way the regeneration circuit is implemented from the old original. The original drawing shows a standard 25K pot for regeneration. Where the revision "A" and later used a reverse audio 50K regeneration pot and revised feedback routing. From the info I have gathered on the net, these early 904A's never self oscillated, and this was implemented on one of the later revisions, perhaps Rev. "B" 12/19/69 or Rev "C" 11/10/70. Too many times I have read some 904A's oscillate while others do not, with one tech claiming the old ones had fallen out of spec. My 1969 904A did not self oscillate untill I did the proceedure outlined in the Norlin schematic with the piggy-back resistor to upgrade. I have never read any early documentation claiming the 904A could be used as an oscillator. So self-oscillation in 904A's may not have occured untill around 1970. Happy accident they implemented?
It is tough to tell what is what on the Norlin schematic as there so were many changes made, but those reference voltages are still the same as on the old 1967 schematic. Might want to look at or temp. remove those two diodes, and check those 10 ohm and 330 ohm values? The negative voltage combined with lack of regeneration seem related. I would also check for those missing decimal points on those Norlin schematics. Then making sure resistors around that circuit are correct like the R41 and R34 being both 3.3K . Some of those schematics are poorly transcribed from the original and can be a source of confusion/errors with blurry values.
I have two other copies of the 904A, with the original blueprint on yellowed paper, the R.A.Moog schematic, plus that blurry Norlin schematic with the calibration instructions. I can E-mail the copies if you need them to compare.

[EricK]

Would that be true in regards to those prototype 904ABC filters that guy had on the IDOW doc?

[BrianK]

The August 1969 built 904A is a revision "A", and it differs in the way the regeneration circuit is implemented from the old original. The original drawing shows a standard 25K pot for regeneration. Where the revision "A" and later used a reverse audio 50K regeneration pot and revised feedback routing. From the info I have gathered on the net, these early 904A's never self oscillated, and this was implemented on one of the later revisions, perhaps Rev. "B" 12/19/69 or Rev "C" 11/10/70. Too many times I have read some 904A's oscillate while others do not, with one tech claiming the old ones had fallen out of spec. My 1969 904A did not self oscillate untill I did the praoceedure outlined in the Norlin schematic with the piggy-back resistor to upgrade. I have never read any early documentation claiming the 904A could be used as an oscillator. So self-oscillation in 904A's may not have occured untill around 1970. Happy accident they implemented?

I have yet to find a pre-70 recording that has good self-oscillation of the LPFilter, listen to Lucky Man (a pre-summer 1969 filter) that doesn't QUITE go into oscillation at the end section. High resonance, yes, but no feedback. This is not proof, but a good example of a time-relevant filter being used with high feedback. I will go test some of the pre-69 filters here and see if anything does oscillate, although I know the summer-1965 module I have here does not.

[CZ Rider]

This is exactly how my stock August 1969 904A filter worked. At a regeneration setting of 10 the resonance was high, but not quite able to produce a sine wave oscillation. Although with a little feedback patching, I was always able to make it oscillate if that was the effect I needed. The 904A inverts the incomming signal, so a negative feedback loop would be all that was needed to make it oscillate. The 904B is the same way, even though there is no regeneration dial. The 904B does not invert the incomming signal, so a positive feedback loop is needed to make it oscillate. The 904B feedback oscillation/resonance patch is one of my favorites now!

The Norlin service manual page for the 904A was a little confusing the way it is written. But after doing the proceedure for selecting, steps 9 and 10 read as an upgrade to make the older revision filters break into oscillation sooner than originally designed. This would explain why some oscillate and some don't at a regeneration setting of around 8 to 10. So right now my summer of '69 filter oscillates and you would have to look at the resistor to see if it has been upgraded. I have no reason to believe this filter was out of spec, but it was designed at that time to only go to the very edge before oscillation. That's my guess anyhow!

[CZ Rider]

Would that be true in regards to those prototype 904ABC filters that guy had on the IDOW doc?

Wow, forgot about the begining of IDOW. Had to go back and watch that again. Although they show the first ever 904 combo in a rack form that had no adjustable resonance. It is unclear when they added the resonance/regeneration dial. Could have been added a few years later or a few months later. I have seen early 1U modules labled 904 low pass with just a frequency and range knob, so that is not the only early one without any regeneration.
Interesting that right after showing that original, they demo a IIIP system with the classic 904A filter sweep. Notice the regeneration knob gets turned to 10 and it is just on the brink of self oscillation. So add in another one that did not oscillate. My guess would be that making that regeneration knob go to eleven would be the Rev. "B" done on 12/19/69. Probably from that point on? This is also around the time of the Minimoog developement, and the Mini filter oscillation was listed as the sixth sound source in the operations manual.
Just thought it was worth mentioning that all these non-oscillating 904A's probably are not broken, just not yet implemented.

[rarecomponent]

Thanks for very useful info and offers of help etc.

So far my 904a#2 works fine....but poor old 904a#1 is still lacking..their both identical in every way .(maybe some different caps)

On my Norlin Schematics R11 value is quoted as 1.R k ..2.2A another copy says ..1.Rk 2.2k...all confusing...

But for the record my first attempt at a 904a _18 years ago..I used my own layout which was pretty rough and Bc109s and BC214ls in place of the specified transistors and NONE matching...this filter used to howl and cry at full resonance with superb filter sweep. I am seriously thinking of reinstalling it...
It was probably just happy accident who knows? But compared to my faithfully copied 904a clone with original tracks /matched trans and vintage components..it puts it in the shade.

[robotmakers]

Just for the internet archives, the Don Martin era 904a is a clone of the Rev "C" 11/10/70 version, and self-oscillates with abandon.

Cheers,
Roger

[BrianK]

Just did a listening test with 8 Moog 904a modules together [borrowed from everyone In town]- and all the 1960s-era ones did not self-oscillate (actually pre-72, the R A Moog versions). What's funny, is that some of them had already been modified, with a jumper resistor to increase the gain in that circuit, BUT still no self-resonant oscillation.

[analogmonster]

Just did a listening test with 8 Moog 904a modules together [borrowed from everyone In town]- and all the 1960s-era ones did not self-oscillate (actually pre-72, the R A Moog versions). What's funny, is that some of them had already been modified, with a jumper resistor to increase the gain in that circuit, BUT still no self-resonant oscillation.

I don't know whether you refer to R10 of the 904a schematic of the service manual, but if not you can increase this resistor to achieve self oscillation, e.g. to 10k or higher.

[rarecomponent]

Thanks Anmon......thats a great tip !

I tried/experimented several value resistors at R10 and at the moment a 1.8k is in there and yes, it certainly has made the filter sit up. That extra 800r at R10 has now created self oscillation and enhanced resonance considerably. Maybe there were some high value tolerance 1k resistors used in some Moog 904a's creating the some do/ some don't self oscillation ?

[CZ Rider]

Here is a photo of my Rev A from drawing #1149 R.A.Moog 904a. I have highlighted the R11(1) with the tagged on selected resistor before soldering in place. (Latter soldered and leads clipped.)

The original placed resistor was a 1.8K. This 904A would not oscillate, but was just on the brink at a regeneration setting of 10. I did exactly what is outlined in the service manual and connected a resistor substitution box arcoss the leads of the 1.8K resistor. As I dialed in and selected different resistors values while observing the regeneration dial response. At a selected 2.2K value I found the best response on the regeneration dial. Oscillation began around 8 and was full at 10. Other lower resistor settings made the 904A oscillate earlier like a setting of 6 or 5. Too much higher than 2.2K and it did not oscillate properly.
The combination of 1.8K and 2.2K in a parallel circuit equals a 990 ohm resistor. So adding in a resistor in parallel lowered the resistance.

I was quite happy with this Moog recommended modification. The original regeneration range from zero to ten, is now in the zero to eight range. And from eight to ten gives oscillation. The oscillator range is not that great. Does not get those ground shaking sine waves some filters get. It is about a three octave range, and a challenge to get the volt per octave range dialed in. Not perfectly linear either, but it does sound sweet.
What does a 904A sound like as an oscillator?
Here is a small 18 second sample that will give an idea of the range and tone when played from a keyboard.
Moog 904A as an oscillator MP3 download 222kb
Never really knew for sure, but thought that classic Tomita whistle sound was done with the 904A slightly modulated, and a liittle bit of white noise. And a bit of 951 glide with massive ammounts of talent.
YouTube link to Tomita playing his whistle. Sure looks like he is playing the 904A. And that range sounds about the same as I am getting where the volume dips at either end of the range.
Tomita YouTube link When you wish upon a Star

[analogmonster]

Thanks Anmon......thats a great tip !

I tried/experimented several value resistors at R10 and at the moment a 1.8k is in there and yes, it certainly has made the filter sit up. That extra 800r at R10 has now created self oscillation and enhanced resonance considerably. Maybe there were some high value tolerance 1k resistors used in some Moog 904a's creating the some do/ some don't self oscillation ?

Yes, I am sure it is a question of tolerance. In my clone the self oscillation works with R10 = 1K, starting around 9. I am quite sure this is because the transistors I use have higher betas than the originals.